Thank you for your comments, I don't have any disagreements with your
assessment. To summarize there are three changes to make
- clean up the log message, make it more concise (in conjunction with
added comments in the code) and adhere to the guidelines.
- improve/add comments in the code
- fix the UTF8 <-> native encoding issue when remembering the
external folder length
... but since I'm on vacation next week and busy getting work done
this week so I can afford to take the vacation, you won't see
anything for a while ;-)
>
> print_status() just calls svn_cl__print_status_xml() and
> svn_cl__print_status() to do the real work. Both of those functions
> take their 'path' arguments in UTF8, as does print_status() itself,
> since it implements a public interface and Subversion public
> interfaces always take paths in UTF8 unless otherwise stated.
>
> So, you can't just print out 'path'; you have to convert it to native
> encoding and print that instead. And this solves the problem that
> back in notify.c, you calculate 'extern_path_prefix_length' based on
> 'path_local', because now the calculation will match what's being
> printed later.
Oops! I missed the distinction between path_local in notify.c and
path in status-cmd.c. That was silly, I should have known the names
were different for a reason.
What is the apr function to convert to UTF8? I don't suppose there
an easy way to fold the conversion into the strndup without
converting the entire path, not just the prefix that I need to
print? I'm just thinking of avoiding more of those allocations, but
I guess if we clear the pool when the message is printed it doesn't
matter. I just didn't have time to understand enough of how the pool
was used to feel comfortable with that. How do we know when it is
"safe" to clear the pool? It seems any user of the pool has to be
very careful that it isn't cleared before they are done with it.
> Have you verified that the change passes 'make check', by the way?
I did verify that the changes passed "make check". There was concern
expressed in the comments for issue #1935 that the tests might fail
after this change, but they did not.
I wrote earlier in this thread:
> The results are in... everything passed but:
>
> FAIL: lt-fs-test 13: delete nodes tree
> At least one test was SKIPPED, checking /Users/scottpalmer/dev/
> svn_test/subversion/tests.log
> SKIP: utf8_tests.py 1: conversion of paths and logs to/from utf8
>
> I'm not sure if the fs-test failure is my fault. It doesn't sound
> like it should be.
Now I'm concerned about the skipped utf8 test... Perhaps it would
have caught my UTF8 strlen mistake?
I'm running on OS X, I think there were some known issues with that
test on OS X at one point.
Thanks,
Scott
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Aug 25 15:42:36 2005